Eureka 427: What role can AI play in healthcare? 

0:14You're listening to Eureka on Monocle Radio, brought to you by the team behind the entrepreneurs, The show all about inspiring people, innovative companies, and fresh ideas shaping global business. I'm Laura Kramer. As the recent Artificial Intelligence Action Summit wrapped in Paris, one key takeaway was the growing need for specialized data sets in sectors that serve the public interest. So what role can AI play in something like healthcare, one of the most vital and human-centered industries? Andres Klevey is the co-founder and CEO of Cordy, a healthcare-focused AI platform designed to support doctors and nurses by reducing paperwork, improving decision-making during patient interactions, and minimizing diagnostic errors.

1:03And hopefully by cutting down on all this admin work, doctors can spend more time with patients, leading to better care for everyone. So how does it work, and what might the future of AI-driven care look like? Here is Andreas in conversation with our very own Tom Edwards. Cordy is an AI foundation model, like the ones most people know living inside ChatTVT. But ours is solely built for healthcare. So it's been trained on millions of patient interactions to become quite good at understanding how to be helpful to a doctor and a patient. Now, I think this is really interesting, Andreas, because I guess one of the things that people worry about a little bit when they talk or hear about the rise of AI-driven techs is that it can have a dehumanising effect.

1:48Obviously, if you think about healthcare, wellbeing, the medical profession, that's an area where people say, look, you know, we cannot lose the humanity. It's fundamental to this business. Talk a little bit about how, I guess, the transformational impact of AI through Corti's prism is actually in service of exactly that pursuit. People are 100 % incorrect about that fear. Tell us how you can rationalize that and explain it? That's a great question. And I actually think the fear of dehumanization through AI is so misplaced because that's exactly what's already happened today, pre-AI. We have doctors and nurses who are, for many cases, very, very stressed, close to burnout, working many hour shifts and ending up doing a third of their time spent on just administration, paperwork, non-care, not the stuff we want them to actually spend their time on, but stuff that they generally shouldn't.

2:39So in the world we live in today, this is already happening. We've massively dehumanized healthcare by putting all these burdens on our providers who are our best and brightest. And I have taken for many of them a Hippocratic oath to care and be there for us on the worst days of our lives. What we're just trying to say is ultimately, if we ever want the care we believe we deserve, then we just need to build way better tools for the people who have devoted their lives to be there for us when we really need it. Well, let's get into the detail then. What are some of these tools, Andreas? And tell us a little bit about how they're driven by AI.

3:10Because there's some areas, if we look at volume of caseload and processing, I think it's easier, almost intuitively, to see the benefits. But tell us about some of the mechanics by which Corti can have this transformational impact. Corti's vision from the get-go was to build. And this is all the way back to when we started Corti in 1617. there was really no JTBTs, no fancy AIs. It was just IBM Watson having just won Jeopardy. That was sort of it. And our vision back then was an AI that would be sort of the silent listener, the invisible secretary that would just be sitting on the sidelines, but being capable of following patient interactions in real time, nodding down everything they hear, taking the notes, writing the emails, following up, doing what's called medical coding.

3:56And all of these things are quite complicated. They're quite hard, but they are so important for making sure that we as patients, we travel through healthcare in the most meaningful way. And some of the cases we see today, many years later, after doing tons of research and science on how to actually deliver this kind of invisible secretary or support is a lot about everything that takes doctors away from the patient and back to paperwork. And that usually is everything that has to do with getting paid, sending the patient on, readmitting the patient, following up on the patient, inviting the patient.

4:32Everything around the engagement of patients is quite broken. It's quite archaic. It's very paper-driven. I think we've all been frustrated by it. And then there is a host of use cases around ensuring that what is then said, shared, and done is to the highest quality. And I think we all know that there are so many bright people in healthcare, but there is also a risk of variance that we as patients might not be thinking a ton about. Just the pure fact that if we go to a doctor in one place versus another place, there might be a difference in outcome. And that's not because one doctor is less smart than the other.

5:04It's just because it's terribly complicated to make sure that it's always the same kind of advice delivered. And that's exactly where AI can come in and help. And it's interesting, you mentioned use cases. And I think, again, to maybe the untrained observer, they've seen a lot of AI startups, probably what, post-2020, riding this kind of big global wave. And maybe the skeptic says, I don't really see the use cases. I'm not sold on this. I think what's fascinating with Corti, Andreas, is that there are so many proven use cases already. And it doesn't matter whether we look in Sweden or in Boston or switching over to the West Coast in Seattle.

5:38Real use cases, there's a demonstrable benefit. That must be so powerful when you're talking to investors, prospective partners, maybe to those sceptics, because you can say, don't just listen to what I'm saying. You don't have to believe the evangelist. Look at the data. We can show this working. That's absolutely true. And especially also to making sure we get the right team assembled. So just this day, today, Cordy will be helping something like 250 ,000 patient interactions somewhere on earth. And from a founder who founded this company with a very, very deep personal experience in healthcare, my mom was in psychiatry and got attacked by a patient that ultimately led to my entire childhood in and out of healthcare institutions, getting the help she needed, and she had a fantastic recovery, but couldn't have done it without the help of excellent healthcare professionals.

6:26And just knowing what it could have meant to many of these people, having a sidekick that could do a lot of the admin, could do quality assurance, could give second opinions, could do all the tedious labor. I can only imagine that the next Andreas and his mom would hopefully get the benefit of a little bit more eye contact, a little bit more time with doctor. And that's super motivating for everyone here. Just on that point, as a founder and as an individual, Andreas, is that really important? Clearly, from what you say, it's baked then into the sort of mission of the business, that experience.

6:53Do you think it makes the company today very different than what it would otherwise have been? Do you take it personally? Do you see that as an advantage, as something that positions the business in the best light to move forwards? Because you take it personally because it's a matter of personal purpose as well as being something that's compelling from a business point of view? Absolutely. And I think sort of as stewards of a new kind of supportive technology, our ultimate goal is to make sure that patients just get more healthcare that is trustworthy at a lower price and easier access. Because we are patients, right?

7:28Every one of the hundreds of Corleans, our patients have families who have intricate relationship with healthcare. And we're all here because we want to give back to all those people who helped us. Ultimately, building this kind of technology is not about sort of riding the wave of Silicon Valley-esque technologies. We've been doing this quite a while longer. It's ultimately to try to build some more trustworthy infrastructure of technology that can humanize that relationship between patient and provider more. And ultimately ensure that more patients just get what they need when they need it and not sort of ultimately make it bound by where they live or who they have access to.

8:00That's interesting. I wonder, has the discourse changed? If we go back, as you said, 2016, 17, getting this business started, very different narrative. And I guess the pandemic played a big role in this as well, Andreas. I imagine you'd agree. In terms of people's perception of talking about their broader health, so not just health care provision, but sometimes slightly more nuanced areas, wellness, well-being. Do you think that Corti can move the needle on all these, I forgive the pun, can be effective addressing all of these? So not just health care provision, driving efficiencies and efficacy in that theatre as well, but actually helping people to feel better?

8:38Or is it kind of too early to talk about ambitions like that? No, I think it's absolutely on point and it's about time. And I'm sure many of the listeners here are stewards of their own health. They have active make decisions about wearables, subscriptions, their diet, how they think about their health generally. And I think it's an entire generation entering the consumption of healthcare and being of age where you start really consuming healthcare that are much more attuned, aware and insightful than ever before. They have access to more things. They're more focused on getting the right thing out of those interactions.

9:11They're not sort of shunning away from healthcare. As former generations more felt, it's sort of like, oh, let's steer clear. I think that's a quite massive change. And I think AI potentially plays a quite vital role because when we've all done our genome and we have tested our microbiome and we have engaged a ton of doctors and we've all found a workout plan and we're totally taking control of our wellness, there's an entirely on top of that of ensuring that you get the right feedback. You can scour for all that data and you can build that ownership into tangible outcomes. And I think AI, again, plays a role there in empowering patients.

9:43And ultimately, the goal is that some of these solutions will help drive consumption. There's more healthcare being used. And I think that's sort of core to our thesis. So today in an OECD, like classic Western country, you have something like seven interactions per person per year on average with healthcare. We believe that in the future, that's going to be way, way, way more. So you're going to have way more interactions through all those devices, through all your services with AI that represents healthcare reasoning that allows you to be much smarter, make much more informed decisions and ultimately make healthcare much more preventative and less sort of reactive.

10:15Well, and I suppose as people age and they'll live longer and be older societally everywhere, particularly, as you said, in sort of Western developed countries, in a sense, that's good because there'll be more end users, right? More consumers that need extended healthcare provision. But there's also a lot more disagreement about how best to pay for, to service that requirement. Given that there is such a, you know, it's quite a dynamic area of discourse. And given that the pace of technological advance is so rapid, how on earth, Andreas, do you and your colleagues, you know, keep an eye? You've got to think about, you know, by 2030, we're going to be living longer, requiring more care.

10:59You've got to factor that into your strategic thinking, right? How do you do that as a leader of a business where even the fundamentals that you're working with, the building blocks, are changing on an almost week-to-week basis? That's very difficult. Are those things in tension? I love the question. I think obviously we can think about it directionally. And I think that the directionality of what we believe will happen ultimately is that if we look at technology generally, what it does really well is deflates cost. So in technology, like back in the day, if we wanted to travel somewhere on Earth, there were very few travels.

11:29It got quite expensive. It was very hard. Today, you can travel to more or less anywhere on earth within 24 hours. And it's relative to average income in Western countries. It's decently cheap. We believe that that deflation hasn't hit healthcare at scale yet, especially not in certain markets like the US. So my point here with being like, you're right, more people will definitely be in need of healthcare. They're probably going to be needing and using healthcare differently. So as I said before, more preventative care and maybe less reactive care, at least hopefully. But ultimately, they're going to be able to consume way more healthcare as AI will drive down the average cost of care.

12:02So if you think about the US healthcare alone, it looks like they're spending something like$4 trillion here a year on healthcare. And roughly a third of that is associated with administration. So the question becomes, A, how much of that administration could be removed, hopefully lowering prices, and B, how much of that care could be converted from reactive care, which is quite expensive, to preventative care. And ultimately all of this will drive down costs and it will give us more access. And that makes me quite excited. Well, channel that excitement for us, Andreas. What are you most excited about?

12:33And again, I guess speaking to that last question, how do you calibrate your expectations? You look at a three, five, 10-year time horizon. Someone from outside says, wow, this sounds really interesting. What Andreas has said resonates with me. I'm going to watch with real interest what happens next. What should we look out for? Is it the next breakthrough that the technology enables? Is it the next market you enter into? What should we be looking out for? and what gets you springing out of bed in the morning with the most excitement? The boring answer, which is true, is it's great partners and great colleagues.

13:02But what we think a lot about and what's happening next is we need to train the AI to be helpful to more people who are facing different challenges so we can remove any kind of risk of bias or over-indexing. And we do that by scale. So we need to build the AI better so it's more helpful, so it can help more people. That means more markets, more opportunities, more specialties. Healthcare has a lot of specialties, from radiology to pathology to many, many more. So what we'll be spending our time on the next many years is just scaling to every corner of earth, especially all the underrepresented languages and opportunities there because patients need it as well.

13:37And then all the specialties we want to be helpful in. And what we're excited about is that we've seen historically that technology has been pushed, especially software on healthcare. And today we start to see a pull. So clinicians want to test more. They want to try out more. They feel they have an avenue to start tackling some of their challenges through AI. So I think a lot of these smart clinicians are going to build fantastic solutions on top of stuff like ours. And that ultimately will be a game changer. Last thought, Andreas. Maybe there's someone listening who's still cynical about AI.

14:07And they say, look, there's too many unanswered questions. And I'm reassured by Andreas's commitment to benchmarking excellence and rigor and eliminating bias. But something about it makes me uncomfortable. What would be your message to them? Is it that they need to interrogate the data more rigorously? Do they need to wait and see? How do you win a round? Or do you just say, look, talk to me in 18 months or two years or three years when I've got more use cases? You must get constantly actually challenged on this. How do you, as someone who is in the vanguard and is so committed about what this can mean in terms of opportunity and accessibility, how do you win that argument?

14:44Or is it about elegantly sidestepping it when it throws itself up? No, no, I think that's the big big of technology companies and founders and startups is to try to evade this entire problem of sort of compliance and bias and alignment. How do we actually make these models deliver on the promise? So remember, we founded Cordy before the chat TVT hype, before the pandemic. And back then, there was very few people doing AI here. And now there's a lot. And what I hear increasingly is like either the big tech wants to regulate or the founders wants us to stay clear. I ultimately think to anybody who's a skeptic out there, I think it's obviously manifold.

15:19One thing is let's stick to the scientific method. let's actually publish research. Let's make sure it's peer reviewed. Let's prove there's alignment. Let's prove we do audits. Vendors who don't do any of these things like publish audits or do alignment of bias audits, they're probably not doing it at all. So I would just say like in healthcare and in critical industries, let's hold the companies like ours to the absolute highest standard as we do the people working there. And I think generally for the general application of AI, I think more and more people see the benefits and I think they'll see more of it.

15:48Keep holding it to a high standard generally on quality. And then I think there is a last part of it, which is just very tied to our individual experience of AI. And I think right now, it may be just important to remember sometimes when we get swept away by the media cycles of the worry about like all being toughness in AI taking our jobs. And right now it's actually just quite nice at writing birthday speeches and helping us out at work. It's not actually replacing us. So maybe we have time to figure it out.

16:21That was Andreas Clevy, the co-founder and CEO of Cordy. You can learn more by heading to cordy.ai. And that's it for this episode of Eureka. Tom Edwards will be back next week and look out for the entrepreneurs available every Wednesday. Eureka was edited by Jack Dewars. Listen again and find out more at monocle.com. That's where you can join the club and subscribe to Monocle Magazine or follow us and catch up with the archive via your preferred podcast platform. To contact the team, email me at lrk at monocle.com. I'm Laura Kramer. Goodbye, and thanks for listening to Eureka.